Fed- zyxwv

thuringiensis for T huring iensinzyxwv

batchCuItureofBaci//us

Production in a Tower Type Bioreactor

Jian-Zhong Jong, Ding-Yu Hsiun, and Wen-Teng Wu*
Department of Chemical Engineering, National Tsing Hua University,
Hsinchu 30043, Taiwan, ROC

Yew-Min Tzeng
Department of Food Engineering, Da- Yeh Institute of Technology,
Dahtsuen, Taiwan, ROC
zyxwvutsrqp
Received September 30, 1994lAccepted June 14, 1995

Fed-batch culture of Bacillus thuringiensis in a modified
airlift reactor has been developed by using adaptive con-
trol of glucose concentration in the reactor. The glucose
concentration was estimated via a correlation equation
between carbon dioxide production rate and glucose
consumption rate. The estimated glucose concentration
as the output variable was fed back to computer for cal-
culation of substrate addition. The modified reactor was
sin production
Fed-batch culture is commonly applied to fermentation
processes. The cultivation is employed for prevention of
substrate or metabolite inhibition. However, fed-batch cul-
ture is more complex to operate than batch or continuous
culture. A proper feed rate of substrate is required during
fed-batch culture. In this study, glucose concentration is
estimated on-line and is taken as an index for substrate
zyxwvuts tio n

an airlift reactor with a net draft tube. The airlift reactor

had high oxygen transfer rate and low shear stress which
were important factors for production of thuringiensin.
Fed-batch culture of Bacillus thuringiensis in the modi-
fied airlift reactor provided significant improvement of
thuringiensin production. 0 1995 John Wiley & Sons, Inc.
sively investigated
addition. Adaptive control of glucose concentration in the
fermentor is applied during fed-batch culture. Some reasons
for using adaptive control are: (1) the analytical model of
the fermentation process is difficult to obtain; and (2) sys-
tem identification can be easily employed during cultiva-
zyxwvutsrq

Key words: Bacillus thuringiensis glucose thuringien-

because fermentation processes are always sluggish.
In the present report, fed-batch culture with adaptive control
of glucose concentration in the modified airlift reactor was
INTRODUCTION studied, which provides a proper cultivation condition for
thuringiensin production.
Thuringiensin has shown efficacious zyxwvuts Thefermen tati on sy stemissho wn in Figu re1.Th efermen

Bacillus thuringiensis has become one of the most exten-

microorganisms for production of bio-
insecticides in recent years.6393’1* 1 2 * 1 5The microorganism
MATERIALS AND METHODS
can produce both an endotoxin and e x o t o ~ i n s . ~ *Thu~~- ‘ ~ ” ~
ringiensin, one of the major exotoxins, is a thermostable
and water-soluble metabolite from cultivation of B . thur- Equipment
ingiensis. control
against flies,3 Colorado potato beetles,* lygus bugs,14 and
tor was an airlift reactor with a net draft tube.I7 The fer-
several species of mites.’
mentor was 13 cm in diameter and 200 cm high. The con-
Providing an appropriate environment for high cell den-
centric draft tube, 6.5 cm in diameter and 100 cm high, was
sity cultivation of B . thuringiensis to produce thuringiensin
is the main task of this study. Jong5 used an agitated fer- a net tube with a mesh number of 24. The reactor which was
surrounded by a jacket for temperature control was made of
mentor, a bubble column, and an airlift reactor with a net
stainless-steel (AISI 304).
draft tube for batch culture of B . thuringiensis. The exper-
The sparger located at the bottom of the reactor was a
imental results indicated that different reactors gave differ-
porous plate with pore size of 40 to 50 km . The DO sensor
ent morphologies of the cells. Shear stress in a bioreactor
(Ingold) and pH sensor (Bradley-James, Model F-600) were
had a significant effect on both growth of B . thuringiensis
positioned at 45 cm and 70 cm from the bottom of the
and production of thuringiensin. Because the airlift reactor
reactor, respectively. A metering pump was implemented
with a net draft tube has a high capability of oxygen trans-
* To zyxwvutsrqpowhomall correspondenceshouldbeaddressedsensorswereinterfacedtoacontrolunit,anIBMPC /AT

for substrate addition during fed-batch culture. The off-gas,

fer,” and low shear stress, fed-batch culture of B . thuring-
after filtration and drying, was analyzed by an oxygen an-
iensis in the airlift reactor is investigated in this study.
alyzer (Teledyne, Model 326) and an infrared carbon diox-
ide analyzer (Fuji, Model Zara 2061). All the analyzers and

Biotechnology and Bioengineering, Vol. 48, Pp. 207-213 (1995)

0 1995 John Wiley & Sons, Inc. CCC 0006-35921951030207-07
Figure zyxwvutsrq 1.Schematicdiagramofthefermentationsystem

zyxwvutsrqThe
with a PC-Lab AD/DA card (Advantech, PCL-812PG, Tai- except for glucose concentration of 20 g/L. All media ex-
wan). cept glucose were sterilized together.

Microorganism and Medium Preculture

organisms were cultivated in 1-L Erlenmeyer flasks
Bacillus thuringiensis subsp. durmstadiensis (HD- 199)
containing 250 mL of medium at 30°C on a rotary shaker
from Dr. de Barjac of Institut Pasteur (Paris, France), was
zyxwvutsrqp

(Hotech) at 200 rpm for 12 h. The medium contained 5 g/L

grown on the medium described by Holmberg et al., with
yeast extract and 8 g/L nutrient broth. The inoculum which
the exception that the carbon source was glucose and the
contained vegetative cells was transferred to the reactor.
nitrogen source was soy flour and (NH,),SO,. The compo-
The volume ratio of inoculum to the initial medium in the
sition of the medium was as follows: glucose 10 g/L, soy
reactor was 1 to 12.
f l o u r 30 g / L , KH,PO, 5 g / L , K,HPO, 5 g/L,
MgSO, . 7H,O 50 mg/L, MnSO, . 4H,O 30 mg/L,
FeSO, * 7H,O 10 mg/L, CaCl, . 7H,O 50 mg/L, and
Assays
NaNH,HPO, . 4H,O 1.5 g/L. The feed medium, which
was utilized for fed-batch culture, was similar to the fer- Glucose concentration was measured using the glucostat
mentation medium except for the amounts of glucose and reagent, with a 505-nm wavelength of a spectrophotometer.
soy flour. Two feed media were used. Medium A contained The analysis method for thuringiensin was modified from
100 g/L of glucose and 150 g/L of soy flour. Medium B the methods of Campbell et al.' and Levinson et al.7 High-
consisted of 200 g/L of glucose, 100 g/L of soy flour, and performance liquid chromotograph (HPLC) was used for
40 g/L of (NH&SO,. For batch culture, the composition of determination of the concentration of thuringiensin. Five-

the medium was also similar to the fermentation medium

208
micron particles (RP-18) in a 3.9-mm-diameter and 15-cm-
BIOTECHNOLOGY AND BIOENGINEERING, VOL. 48, NO. 3, NOVEMBER 5, 1995
z
length end- zyxwvutsr cappedcolumn(Merck)wereused.ThemobileThegoverningequationforadaptivecontrolofsubstrate

phase, 50 mM KH2P0, solution (pH 2.8, adjusted using addition during fed-batch culture was:
phosphoric acid), was kept at the flow rate of 2 mL/min.
Thuringiensin powder (kindly provided by Dr
zyxwvut .+.
.deBarjac)dG(t)
--
-a G(t) b u(t)
dt
was dissolved and adjusted to different concentrations to
establish a standard curve of thuringiensin concentration.
where ( ) was the feeding rate and a and b were estimated
The numbers of spores and vegetative cells were counted
by using weighted moving identification with an adjustable
directly by using a microscope (Olympus, BHZRFC, Ja-
identification interval. l5 By using the model-following
pan) with a counting chamber (Erma, Japan).
strategy, a reference model was given by:
dG*(t)
STATE ESTIMATION AND FED-BATCHCULTURE
--
-A . [G*(t)- Gs] (7)
dt
For fed-batch culture, substrate addition was applied to control where G*(t)was the reference value of G(t),G, the setpoint
the glucose concentration at a given value. Glucose of G ( t ) , and A a negative constant. An error signal was
concentration was estimated by using an empirical correlation defined as:
equation between the consumed glucose and integration of
carbon dioxide production rate (CPR). Oxygen uptake rate E ( t ) = G ( t ) - G*(t)
(OUR) and CPR were determined by analyzing inlet and outlet With Eqs. (6), (7), and (8), we obtained:
gases of the fermentor. They were defined as:
a t )

+ ( a - A) . G(t) + b . u(t) + A . Gs
--

- A . E(t)
dt
(9
and V zyxwvutsrqpo wasthe 11

By using the control policy of making the error signal ap-

proach zero,

where C, was the mole fraction of oxygen, C, was the mole

the feeding rate was determined as:
fraction of carbon dioxide, Finwas the inlet air flow rate,
liquid volume in the fermentor. The sub-
scripts “in” and “out” denoted the inlet and outlet, re- u(t) = - - . ( U - A) G(t) - - A . Gs
* * (11)
b b
spectively.
The empirical correlation was a simple linear equation. It
was inappropriate to use the equation for estimation of the
RESULTS AND DISCUSSION
entire cultivation. The cultivation was divided into two
zyxwvu
Eq. (1with A3.Attheendoffed-

stages, the initial stage and the fed-batch culture stage. In A fed-batch culture of B . thuringiensis was carried out in
the airlift reactor with a net draft tube. The initial working
the initial stage, the consumed glucose Gcons(t),was ex-
volume was 13 L. When the estimated glucose concentra-
pressed as:
tion was below the setpoint of 5 g/L, substrate (the feed
Go,,(t) =k, . 44VJCPR dt (3) medium A) was added. The feeding strategy was based on
1) = - batch culture, the
1.6506 and k, 1.4859. The glu cose concentration, G zyxwvuts ()

In the fed-batch culture stage, Gcons(t) became: working volume was 16.8 L. The time course of the culture
is shown in Figure 2. The pH value could be well controlled
Go&) = k, . 44VJCPR dt (4) around 7 by using NaOH (6N) and H3P0, (6N). Spores
produced when cells died out. Thuringiensin was produced
where k , and k, were determined by using the method of almost at the beginning of cell growth. The yield in terms of
least squares. The values of k , and k, were obtained as k , = grams of thuringiensin produced per gram of glucose con-
G(t)zyxwvut[GiV;Gcons(t)+Gf.VJV

= sumed was 0.0639.

in the fermentor was expressed as: A batch culture was also carried out for comparison. The
time course of the culture is shown in Figure 3. The work-
 = X - (5)
where Giwas the initial glucose concentration in the fer-
mentor, V; the initial working volume, Gf the glucose con-
centration of the feed medium, and V, the added volume of
substrate during fed-batch culture.
ing volume was 13 L. Because there was no substrate ad-
dition for batch culture, the initial substrate concentration
was at 19.6 g/L. From Figures 2 and 3 , it shows that fed-
batch culture could produce more thuringiensin. The yield
of batch culture was 0.027 1.
From Figure 2, it was found that, during fed-batch, both

JONG ET AL.: B. THURINGIENSIS CULTURE FOR THURlNGlENSlN PRODUCTION 209

 5 3 -
.
.,
.
.. . ! ..
.
...

..

.
.....
.

...
.

.
.

.
.

.
..

..

. .
.

.. ..
..

.
.
................

DO
z
. . _.:--. - 8

;5-. -; .
,I .
. .

c
t
- . . ... .-* . . . -0

.. ..
.,a I
h 4 - . :. *'
.
' ' ' '

E, I' ... ..
,
. . pH
a
0 . , .. - 6
. ..: . .
' . . . .
.
2 - . . . . . . .. .
*:
..

. .. . .
.
.* ;.. ....
. .

. . . .
. . . . . .

c
. .

I -'. I
zyx
'

..
"
. .
1111 ,111 " I:
7

1.6x10'o - - -Thuringiensin6

0.0
-
zyxwvu010 Time(hr)304

1 1.4~10'~ -5 2
a - 5
- 4 p.5'
. 3
E -

9.

2.

b 8 . 0~10~- &r< 3 3
!f

t
0

$ 6.0~10-~
a ;2
f 4 . 0~10~-
0

\ - 1
2.0~10~ -
Cell .
 Figure 2. Time course of the fed-batch culture with the feed medium A

OUR and CPR had maximum values. The activity of cell
growth reached a maximum value in a short period of time
and then decreased even with sufficient supply of carbon
source. The only main nitrogen source from soy flour might
be unsuitable. Feed medium A was modified by adding
ammonia sulfate as feed medium B, which was applied for
another fed-batch culture. The cultivation conditions were
zyx
the same as those of the first fed-batch culture. With feed medium B, the
cultivation had the problem ofdissolved oxygen
limitation. To sufficiently supply the dissolved oxygen for cultivation, the
air flow rate was varied from 2.5 VVM to 4.0 VVM by using computer
control. The variation of air flow rate was based on the following steps: (i)
if the value of DO was less than 1 PPM for 3 min (the sampling

210 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 48, NO. 3, NOVEMBER 5, 1995

 zyxwvutsr
I
50 ----- 1 J

. . -

-
. .
-
1 1 1 1 . .
;"&.&&+.i
1 1 1 1

....
10
1 1 1 1 1 1 1 1

............... - 9
DO

- 8
zyxwvuts ..

. . . I
. . .
.
.
. . . . . . . 7
PH

2 i5
1

8.0~10 ~ - 20 2.5
zyxwvutsrqponmlk1

'1
18
7.0~10~ -
2.0
16
Thuringiensin
14 $
5.
td

h zyxwv 3

-I
m

10 g.
E v

8 4 2
h

8 1.0
?i. 0

g 3.0~10~- 3 Spore
8 a 6
a
td 2 . 0 ~ 1 0 ~
-
t
- 0.5
4
8
1.0~10~ - 2
 0.0- 0 0.0
0 10 20 30 40 50
Time (hr)

Figure 3. Time course of the batch culture.

time was 1 min), the air flow rate increased 0.5 VVM until
it reached 4.0 VVM; and (ii) if the value of DO was higher
than 1 PPM for 5 min, the air flow rate decreased 0.5 VVM
until it reached 2.5 VVM. The time course of the cultivation
is shown in Figure 4. The activity of cell growth could last
longer as shown from the curves of OUR and CPR. At the
end of the cultivation, the liquid volume was 18 L and the
concentration of thuringiensin was 11.71 g/L. The yield
was 0.0905. Production of thuringiensin was significantly
improved.

JONG ET AL.: B. THURINGIENSIS CULTURE FOR THURlNGlENSlN PRODUCTION 21 1

 10011111 , I111

10 zy
9

-0
I
7

6
I
zyxwvuts

zy
5

3
9 4.0x10'0 zyxwvutsr
-
:
zyxwvutsrqpo
5.0~10'~ Thuringiensin 14

o(l 1 3.0x10'0

- zyxwvutsrqponmlkj
3 ;
E
3 2.0x10'0 -
0 l.oxlo'o -
3. +
p!

2i Spore
VJ
/
Od

0.0-
0 10 20 30 40 50
Time (hr)

Figure 4. Time course of the fed-batch culture with the feed medium B.

stages-theinitial
CONCLUSION
zyxwvutsrqp Hence, the correlation equation could provide a good esti-
mate.
Simple correlation equations were developed for on-line The oxygen requirement for cultivation of B . thuringien-
estimation of glucose concentration in the fermentor. The sis for thuringiensin production was high. Fermentation was
cultivation of B . thuringiensis was divided into two appropriately carried out in the airlift reactor with a net draft
stage and the fed-batch culture stage. tube.

21 2 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 48, NO. 3, NOVEMBER 5, 1995

 The fed-batch culture with zyxwvuts modifiedfeedmediumpro-5.Jong,J.2.1994.Applicationofanairliftreactorwithanetdrafttube

h thuringiensin concentration which was good in production of thuringiensin, Ph.D. thesis, Tsing Hua University, Hsinchu, Taiwan.
for downstream processing.
6 . Lambert, B., Perferoen, M. 1992. Insecticidal promise of Bacillus
fhuringiensis. Biosciences 42: 1 12-121.
7. Levinson, B., Kasyan, K . J . , Chiu, S . S . 1990. Identification of
NOMENCLATURE
p-exotoxin production, plasmids encoding p-exotoxin and a new ex-
molefraction of oxygen otoxin in Bacillus fhuringiensis by HPLC. 172: 3172-3179.
mole fraction of carbon dioxide 8. Paige, M. R., Cooper, R. D. 1990. Scale-up of beta exotoxin pro-
zyxwvutsrqpon 15

error signal function duction in fed-batch Bacillus thuringiensis fermentation. European

flow rate of inlet air Congress Biotechnology, 5th Meeting, pp. 14&159.
glucose concentration in the fermentor 9 . Rodriguezpadilla, C., Galanwong, L . , de Barjac, H. 1990. Bacillus
the consumed glucose thuringiensis neoleonensis serotype H-24, a new subsp. which pro-
glucose concentration of the feed medium duces a triangular crystal. J. Invertebrate Pathol. 56: 28G282.
initial glucose concentration in the fermentor 10. Rossa, C. A , , Mignone, C. 1993. S-Endotoxin activity and spore
the setpoint of G(t) production in batch and fed-batch cultures of Bacillus rhuringiensis.
the reference value of G(t) Biotechnol. Lett. 295-300.
1. Campbell, D
zyxwvutsrq
.P.,

the feeding rate 11. Rowe, G . E., Margaritis, A. 1987. Bioprocess developments in the
total working volume in the fermentor production of bioinsecticides by Bacillus fhuringiensis. CRC Crit.
the added volume of substrate during fed-batch culture Rev. Biotechnol. 6: 87-126.
Ch em. zyxwvutsrqp
zyxwvutsrqpon 3 5:15 61 58 . K.,D.M.

initial working volume

12. Sebesta, K., Farkas, J ., Horska, K . 1981. Thuringiensin, the -exo-
toxin of Bacillus fhuringiensis, pp. 249-281. In: Microbial control of
pests and plants diseases 197G1980. Academic Press, London.
References
13. Sikdar, D. P., Majumdar, K., Majumdar, S. K . 1991. Effect of min-
erals on the production of the delta endotoxin by Bacillus thuringien-
Dieball, D. E . , Brackett, J. M . 1987. Rapid HPLC
-exotoxin of Bacillus fhuringiensis. J . Agric. Food sis subsp. israelensis. Biotechnol. Lett. 13: 51 1-514.
assay for the
14. Tanigoshi, L. Mayer, F., Babcock, H. 1990. Efficacy of
the p-exotoxin of Bacillus thuringiensis to Lygus hespems. J. E o n .
2. Cantwell, G . E . , Dougherty, E., Cantelo, W. W. 1983. Activity of
the $-exotoxin of Bacillus rhuringiensis var. thuringiensis against the Entomol. 83: 220G2206.

Colorado potato beetle and the Ames test. Environ. Entomol. 12: 15. Wu, M. M., Tzeng, Y. M., Hsu, J. H. 1993. A study on high-yield
1424-1427. fermentation of thuringiensin formation monitored by HPLC method.
technol. Bioeng. 22: 1707-1724. zyxwvutsrq Ferment.Bioeng. 70:359-361.

3 . Carlberg, G . , Lindstrom, R. 1987. Testing fly resistance to thuring- J. Technol. 8: 223-230.

iensin produced by Bacillus rhuringiensis, serotype H -I. J . Inverte- 16. Wu, W. T . , Ou, W. H., Chen, K. C. 1988. On-line identification
brate Pathol. 49: 194197 . with an adjustable estimation interval. Int. J . Syst. Sci. 19:
4. Holmberg, A , , Sievanen, Carlberg, G . 1980. Fermentation of Bacil- 1955-1967.
lus thuringiensis for exotoxin production process analysis study. Bio- 17. Wu, W. T . , Wu, J. Y. 1990. Airlift reactor with net draught tube. J .
JONG ET AL.: B. THURINGIENSIS CULTURE FOR THURlNGlENSlN PRODUCTION 213